YouTube yesterday announced a set of updates to its generative AI disclosure framework, moving AI content labels to more prominent locations on both Shorts and long-form videos, expanding its automatic detection systems, and reinforcing the consequences for creators who skip the disclosure process. The changes, communicated by Carlos - TeamYouTube in a Community post, take effect starting May 2026 and build on a policy architecture the platform has been developing for two years.

What is changing and where labels will appear

The most immediately visible change is positional. According to YouTube, starting in May 2026, disclosures for generative AI content will be relocated to more prominent spots on the platform's two main content surfaces.

For Shorts - the vertical short-form format that competes with TikTok and Instagram Reels - the AI marker will now appear as a direct overlay on the video itself. Previously, disclosure labels were placed further down in the interface, where a viewer scrolling through the Shorts feed might not register them at all. The overlay placement puts the label in the path of the viewer's eye before they can watch the clip.

For regular long-form videos, the label will be positioned directly below the video player, above the expanded description. According to YouTube, this is intended to provide context before a viewer reads supplemental information about the content. The previous placement was inside the expanded description, a section that many users never open.

These are not cosmetic tweaks. Placement determines whether transparency measures actually reach viewers or remain buried in layers of the interface. A label that appears inside an expanded description, which a viewer must actively tap or click to reveal, functions very differently from a label that appears between the player and the first line of text.

The two-year road to mandatory disclosure

YouTube's AI disclosure program did not appear overnight. According to YouTube's community announcement, the platform introduced the disclosure process two years ago and has been incrementally expanding it since.

The mandatory AI content disclosure requirements that became enforceable in May 2025 required creators to label content that is meaningfully altered or synthetically generated when it appears realistic. The policy took formal effect on May 21, 2025, giving the platform roughly a year to build creator familiarity before tightening the label's visibility in May 2026.

The two-year trajectory reflects a deliberate approach: establish the rule, embed it into the upload workflow, then adjust the presentation layer to make the disclosures more useful to the people they are meant to serve - viewers.

What creators must disclose and what they do not

The disclosure requirement applies to a defined category of content. According to YouTube's Help Center documentation, creators must disclose AI-generated or meaningfully altered content that:

  • Makes a real person appear to say or do something they did not do
  • Alters footage of a real event or place
  • Generates a realistic scene that did not actually occur

The scope covers content that is fully or partially altered or created using AI audio, video, image creation, or editing tools. It explicitly excludes minor edits. According to YouTube, creators do not need to disclose aesthetic touches like beauty filters or colour correction, AI-assisted production tools such as script generation or rough cuts, or non-realistic content like animation or fantasy sequences.

The boundary matters because the industry has spent considerable effort debating exactly where AI assistance ends and AI generation begins. YouTube's framing pins the requirement to realism and meaningful alteration - not to the presence of any AI involvement whatsoever. A creator who uses a large language model to draft a script, then films themselves delivering it on camera, does not need to add an AI disclosure label.

Creator-led disclosure and the upload flow

The primary mechanism for disclosure remains the "AI use" questionnaire inside YouTube Studio. According to the documentation, during the upload process creators encounter an "AI use" field in the Attributes section. Selecting "Yes" triggers the label. Selecting "No" records that the creator asserts the content does not meet the disclosure threshold.

The process works the same way on desktop and mobile devices. The three-step sequence - go to YouTube Studio, follow the upload steps, then locate "AI use" under Attributes - is consistent across platforms, though the interface labels differ slightly: the computer interface uses "select," while the Android interface uses "tap."

Existing published content is not locked. According to YouTube, creators can update their AI disclosure status on most already-published videos by navigating to the Content tab in YouTube Studio, selecting the video, and adjusting the AI use questionnaire in the Attributes section. This retroactive capability gives creators a route to correct past omissions without requiring a full re-upload.

Automatic detection: when YouTube applies the label itself

A significant part of the updated framework addresses what happens when a creator does not respond to the AI use question during upload. According to YouTube, if a creator does not specify whether realistic AI was used, expanded detection systems may apply a marker automatically.

YouTube's automatic detection applies labels in three circumstances:

Content made using YouTube's own GenAI tools - Tools like Veo or the Fantasia Canvas (known in Portuguese documentation as "Tela Fantástica") automatically carry a disclosure label. These labels cannot be removed, because the platform itself generated the content.

Content containing C2PA metadata - The Coalition for Content Provenance and Authenticity standard embeds cryptographically signed provenance information in digital assets. When a piece of content carries C2PA metadata confirming AI generation, YouTube's systems recognise it and label accordingly. PPC Land reported in December 2025that Google integrated SynthID watermark verification into its Gemini app, and the May 2026 Google I/O announcementsnoted that all content created with Gemini Omni remixing carries automatic SynthID and C2PA watermarks.

Content detected by internal systems - YouTube's own classifiers may flag content as AI-generated or AI-altered, triggering an automatic label even without the metadata layer.

When automatic detection applies a label, creators retain some recourse. According to YouTube, they can correct what they believe to be an error by selecting "No" in the AI disclosure survey under Attributes in YouTube Studio. However, three categories of content cannot be adjusted: content made with YouTube's own AI tools, content with verified C2PA metadata, and content that has been manually reviewed and labelled by YouTube's Trust and Safety teams.

That last category - manual review - is notable. It implies that human reviewers at YouTube are making determinations about AI content in at least some cases, and that those determinations are treated as final.

Discovery, recommendations, and whether labels affect performance

One of the most consequential questions for creators is whether an AI disclosure label reduces a video's reach. According to YouTube's documentation, applying a label does not itself affect a video's recommendations or its eligibility for monetisation.

The platform's search and discovery system evaluates content based on viewer signals - what viewers search for, choose to watch, skip, and choose to interact with after watching. According to YouTube's Search and Discovery help documentation, Shorts are ranked based on performance and viewer personalisation. The system draws on watch and search history, topics and themes the viewer engages with, and engagement metrics including average view duration, average percentage viewed, likes, dislikes, and post-watch survey results.

The documentation is explicit: a disclosure label by itself will not change where a video ranks. Whether viewers respond differently to labelled content - skipping it more readily or watching it with altered expectations - is a separate question that the algorithm tracks through the same engagement signals it always has. If viewers systematically disengage from AI-labelled content, the system will surface it less. But that outcome, if it occurs, would be driven by viewer behaviour rather than a platform-imposed penalty.

PPC Land reported in December 2025 that YouTube's recommendation systems already underwent substantial AI-powered refinements throughout 2025, with Google's Chief Business Officer Philipp Schindler noting during Q3 2025 earnings that "our recommendation systems are driving robust watch time growth in our key monetisation areas like Shorts and Living Room." The home feed shifts documented in late 2025 also demonstrated how platform-side infrastructure changes can alter content distribution patterns significantly, separately from anything creator-controlled.

What happens if creators do not disclose

Non-disclosure carries real consequences. According to YouTube, creators who consistently choose not to disclose may face manual label application, penalties including content removal, or suspension from the YouTube Partner Program.

The YouTube Partner Program - the monetisation framework through which creators earn revenue from advertising, channel memberships, and other income streams - represents a significant stake. As PPC Land reported, the platform now supports 3 million monetised channels globally. Suspension from the programme would cut off advertising revenue and access to other monetisation features.

The penalty structure is not new in principle. YouTube's July 2025 clarification on inauthentic content had already established that the platform's rules apply to all content, including AI-generated material. What the May 2026 update adds is a reinforced public statement of those consequences, alongside the expanded automatic detection that makes evasion harder.

It is worth distinguishing what the platform means by "consistently choose not to disclose." YouTube does not state that a single omission results in a channel strike. The language implies a pattern of behaviour - an ongoing refusal to engage with the disclosure framework rather than an isolated upload error.

The C2PA and SynthID technical layer

Beneath the creator-facing interface lies a technical infrastructure designed to authenticate AI-generated content independent of creator cooperation. C2PA - the Coalition for Content Provenance and Authenticity - is an industry standard that attaches cryptographically signed metadata manifests to digital assets, recording their creation and editing history. SynthID is Google DeepMind's imperceptible watermarking technology, which embeds origin signals into AI-generated content during creation, persisting through subsequent editing and reformatting.

Both systems serve a similar purpose: establishing content provenance without relying solely on a creator's self-reported disclosure. When content carries C2PA metadata or SynthID watermarks, YouTube's systems can detect AI origin regardless of what the creator selected in the upload questionnaire. According to YouTube's policy, labels derived from these technical signals are non-adjustable.

PPC Land's coverage of the May 2026 Google I/O announcements noted that all content generated with Gemini Omni remixing tools automatically carries both SynthID and C2PA watermarks - meaning that creators who use YouTube's own AI tools to generate content will always have a label applied, with no opt-out available.

The practical implication is a layered system: creator self-disclosure is the first line, platform detection is the second, and cryptographic provenance is the third. The May 2026 updates reinforce the second layer specifically, through the expanded automatic detection systems.

Why this matters for advertising and marketing

The advertising community has a direct stake in how AI content is labelled on YouTube. As PPC Land reported in December 2025, all AI-edited images carry disclosure labels, which may influence advertiser preferences for authentic versus AI-generated visual content in partnership campaigns.

Brand safety frameworks at many agencies treat AI-generated content differently from camera-captured content, particularly in sensitive categories. As AI disclosure labels become more visible - moving from the expanded description to a direct overlay on Shorts or a prominent position below the video player on long-form content - advertisers will have clearer sight lines into the nature of the content adjacent to their placements.

The growth of AI-generated content on the platform has already raised questions among advertisers about content quality and authenticity. A more visible and technically enforced disclosure system gives programmatic buyers and direct partners a more reliable signal, though the reliability still depends on the accuracy of YouTube's automatic detection systems and on creator compliance with the self-disclosure requirement.

YouTube has been clear that the labels themselves do not affect monetisation eligibility. But the downstream effects on advertiser behaviour - whether brands choose to exclude labelled inventory at the campaign level - remain an open question that will be answered by market activity rather than platform policy.

Timeline

Summary

Who: YouTube, through a TeamYouTube community post by Community Manager Carlos, addressing all creators who upload content to the platform globally.

What: YouTube is updating its generative AI disclosure framework in three ways - relocating AI content labels to more visible positions (direct overlay on Shorts, immediately below the player on long-form videos), expanding automatic detection systems to apply labels when creators do not complete the AI use questionnaire, and reinforcing penalties for creators who consistently refuse to disclose, which may include content removal or suspension from the YouTube Partner Program.

When: The label repositioning and related updates take effect starting May 2026, building on mandatory disclosure requirements that became enforceable in May 2025 and a framework the platform has been developing for two years.

Where: The changes apply globally across YouTube's desktop and mobile platforms, affecting all creators who upload content to the service. The label positioning changes affect the Shorts feed and the long-form video player interface.

Why: According to YouTube, its community has indicated that viewers want greater transparency about how creators use generative AI in content production. The updated label placement is designed to ensure that disclosures are visible to viewers before they watch content, rather than buried in interface layers that require active interaction to reveal. The expansion of automatic detection addresses cases where creators do not complete the disclosure step voluntarily, aiming to maintain platform reliability for both viewers and creators.

Share this article
The link has been copied!